Autonomous pilotage.
OURANOS is our reinforcement learning artilect for autonomous drone pilotage: trained in virtual environments and deployed in the field via Vertex.
From sim to sky
OURANOS is the robotic AI trained via reinforcement learning in virtual environments, specialised in autonomous drone piloting. It learns navigation, obstacle avoidance, and mission-specific behaviours in high-fidelity simulation before transfer to real quadcopters.
Piloting is autonomous by default; drones can be remotely controlled when needed. The policy runs on Vertex edge hardware and receives commands from the central orchestration layer, with telemetry and video relayed via ground docking stations.
When fixed cameras detect smoke or flames, an alert is generated on the Vertex app and a drone can be deployed to confirm the event; once validated, the system supports escalation to emergency services.
Capabilities
Simulation-first training and robust deployment for reliable autonomous operations.
Simulation training
RL | Custom envs
Policies are trained in configurable simulators with realistic physics and sensor noise to improve transfer to the field.
Autonomous navigation
Path planning | Obstacle avoidance
On-board inference for real-time waypoint following and dynamic obstacle avoidance in unstructured environments.
Fire confirmation workflow
Vertex | DEMETER
When cameras detect smoke or flames, an alert is generated on the Vertex app. A drone is deployed to confirm the event; once validated, the system supports contacting authorities who then alert emergency services.
Fleet coordination
Vertex | GDS | Telemetry
Integrates with Vertex and ground docking stations for mission assignment, geofencing, charging, and multi-agent coordination. Video and metrics are relayed in real time to the platform.